Feature selection technique is a technique to reduce data dimensions which are widely used to find the set of features that best represent data. One area of science that often applies this technique is bioinformatics. An example of its application is the selection of significant proteins in the case of Down syndrome. To find out the most influential protein, experiments were carried out on normal mice with trisomy rats (down syndrome mice) totaling 1080 sample and obtained 77 levels of protein expression. The analysis carried out was divided into three groups. Each group was searched for the most influential proteins using genetic algorithms with fitness calculations using random forest algorithms. The results of the protein selection of the three data groups indicate the relationship of the selected proteins to the improvement of learning ability and memory. The results of evaluating selected protein models show a high degree of accuracy, which is above 98.7% for each data group.